Serious scientific simulation features with Newton?

A place to discuss everything related to Newton Dynamics.

Moderators: Sascha Willems, walaber

Serious scientific simulation features with Newton?

Postby Leadwerks » Thu Oct 31, 2019 2:51 pm

I work with some customers that do some very advanced simulation work, and I was just wondering if there are physics features that you think could be made real-time or nearly so? Things like fracture analysis or fluid mechanics. It doesn't matter if it requires a 32 core CPU.

Any ideas? I am just interested in staying ahead of the curve.
User avatar
Leadwerks
 
Posts: 569
Joined: Fri Oct 27, 2006 2:54 pm

Re: Serious scientific simulation features with Newton?

Postby Dave Gravel » Fri Nov 01, 2019 11:05 pm

Hi Leadwerks,

I'm not sure what is the real question.

About fracture:
Newton have already some features to make precalculated fracture.
This is only some usefull tool functions that help the users to create quick precalculated fracture.
The fracture maths and mesh creation is not really related to the physics engine in self or newton law.
It only become related to newton when you need to generate dynamics bodies from it.
The precalculated fracture is a good exemple for show how to create multiple bodies in one pass update.

About fluid:
Newton have already some features for simple buoyancy exemple.
It is not fluid mechanics but it show already how you can control a group of objects with force or velocity and more..

What I'm trying to say,
Normally what you are requesting here is all external features that do not really come from the main physics engine.
It is possible to have a connection with the physics engine for do it, But many parts is not coming from the physics engine in self.
Exemple for the fluid mechanics simulation, It request special maths and good knowledge about the topic.
The fluid simulation in self is similar to recreate a other physics engine just for the fluid.
It is similar to create a visual particle system.
Exemple you can make a particle system and connected it with newton,
For exemple to deal with some collisions to make reaction like smoke that hit a wall or bouncing particles.

I work with some customers that do some very advanced simulation work.


Normally if you have users that doing very advanced simulation they are supposed to have a good idea about how fluid or fracture system working.
For work on stuff like this the users need to have some knowledge about the topic and the math to use, Plus good creativity and imagination and programming skills.
If you take a look around with other physics solution you can see,
Most of them have multiple system for exemple the physics,car,particle,fracture,fluid,softbodies...

Newton can idle very high object count and the number of core is not the most important for this.
It depending more on the system memory and the video card and the render implementation.
Newton don't run on gpu, Maybe it run a bit slower with very high object count but it stay very stable anyway.
With the gpu it is not always the paradise and it have multiple limitations too on the way to use it and more.

If you don't have problem with the time to render the simulation newton can idle a lot objects.
For exemple the blender3D simulation that you can find in youtube with very high stack object count,
Most of them is not in real-time.
It is prerendered and presimulated and it take times and after it is played like a video.

Off-topic:
I have take a quick look in your forum to see what is going on by curiosity.
Many users trying to compare blender stack simulation with newton stack.
This users try to compare a precalculated simulation with a real-time simulation :/
Newton can do very big precalculated simulation too if you record and just play it like a video.
I don't see any point to try to compare a simulation real-time with something precalculated and prerendered.

I have see some topics in your forum about the vehicle too.
You saying that you are working for make a vehicle system.
After you say that you use the newton vehicle system.
And after you say all is broken and you can't make vehicle anymore.
I am mixed... You have create a vehicle system or you use the old newton vehicle system ?
You say to all users newton can't support vehicle ??? I have read correctly ?

First of all the vehicle is more like a construction feature make from the user and not really a engine feature.
It is nice to have a engine having the feature directly implemented, But in general it is separate from the engine.
Again if you take a look on other physics solution you can see most of them use separate implementation.
Many do it like a plugin and don't share the source code.
Most of the time the source code is not share for only one reason, Because it is full of hack to make it work.
It work but with multiple problems that you need to deal with it because you can't change anything in the original code.

Newton is not broken for make vehicle, The vehicle have always work with newton.
Plus the vehicle code is not hide to the users and any users can change the code or help to fix things...
If you try to make a multibody vehicle, Yes it can have some trouble and maybe the tire can bounce over some terrain edges.
I have already make a quick tuto in newton sdk about the multibody vehicle.
And Julio have already give some good solutions for the tire bouncing at this time.
If I remember I have see only one or two users trying or asking question about it.

It is very cool that Julio take times to try to implement feature like vehicle in the engine.
It is even more cool when the users following and try to help or give suggestions.

What I don't understand you say the vehicle is broken to all users but you don't share the code for try to make it better.
If i'm not wrong at this time you are using a vehicle implementation already public in the sdk.
Why not sharing it and try to make it become better with the users, You have write many personal or private parts in it ?
And after you say newton is broken ? I just try to understand the point to say it to all users.

Just for fun try to make a real multibody vehicle with other physics systems.
If you never have see a vehicle behaving like exploding pop-corn you risk to have some fun with the result..

Plus the vehicle raycast have always work pretty good from newton too.
It is pretty easy to implement and it don't give any trouble over terrain.

I'm not here to promote anythings and I don't make money from it.
You can see in my youtube video I never have get trouble to make vehicle with newton,
Or to have high objects count.
I'm not here to make a judgement, I just try to understand.
Sorry for my english and if I boring you with my questions.
You search a nice physics solution, if you can read this message you're at the good place :wink:
OrionX3D Projects & Demos:
https://orionx3d.sytes.net
https://www.facebook.com/dave.gravel1
https://www.youtube.com/user/EvadLevarg/videos
User avatar
Dave Gravel
 
Posts: 800
Joined: Sat Apr 01, 2006 9:31 pm
Location: Quebec in Canada.

Re: Serious scientific simulation features with Newton?

Postby Leadwerks » Sat Nov 02, 2019 12:09 pm

I am just thinking of ideas for the future that might be useful for aerospace customers. We have 32 core CPUs now and I imagine things that used to not be possible in real-time will be possible in the next few years.
User avatar
Leadwerks
 
Posts: 569
Joined: Fri Oct 27, 2006 2:54 pm

Re: Serious scientific simulation features with Newton?

Postby Dave Gravel » Sat Nov 02, 2019 3:07 pm

We have 32 core CPUs

The 32 cores helps for sure but it is not only a core historia.
Many things need to become take in count.
It depend a lot on what you try to achieve.
From my tests the render pass optimization is very important too when you try to have very high object count.
It's important to use the less possible the cpu from other things,
Or use it very efficiently.

About the old newton vehicle component,
Yes it don't have stay always up-to-date with the engine update.
It is only the component, Newton have always work good for make vehicle.
I have always success to get the vehicle working from multiple implementation methods.
And i'm really not the most advanced user about vehicle maths.

The multibodies have always give more trouble over the terrain but the problem is very little.
I don't know many engine that can support vehicle joints construction like the multibodies + nice tire shape collision.
Like I have say Julio have always give good configurable solutions for fix the bounce problem with the terrain.

Edited:
I talk a lot about multibodies vehicle, But it is not the only method for make vehicle from newton.
It is possible to use raycast and the new one is a single body version.

The vehicle have always need some adjustements and it is normal, But nothing is broken like I have read somewhere...
You search a nice physics solution, if you can read this message you're at the good place :wink:
OrionX3D Projects & Demos:
https://orionx3d.sytes.net
https://www.facebook.com/dave.gravel1
https://www.youtube.com/user/EvadLevarg/videos
User avatar
Dave Gravel
 
Posts: 800
Joined: Sat Apr 01, 2006 9:31 pm
Location: Quebec in Canada.


Return to General Discussion

Who is online

Users browsing this forum: No registered users and 16 guests